The results of an analysis of the errors made in sampling measurements of the Land-Mobile Radio Service channel occupancies by the mobile spectrum-occupancy monitors operated by the Federal Communications Commission are presented. Only the measurement errors caused by the limited observation interval (5 min out of a longer time period for each land-mobile channel) are considered in this paper. The major findings were: 1) That the normalized monitor occupancy-measurement error (obtained by dividing the monitor occupancy error by the respective channel occupancy) is larger for low-occupancy channels than for high-occupancy channels. 2) Monitor occupancy-measurement errors increase as the ratio of the 5-min observation time to the total measurement time decreases. 3) Larger relative occupancy-measurement errors occur in measuring 90-percent occupancy with the monitor than in measuring average occupancy. The 90-percent-occupancy value is defined as that value exceeded by only 10 percent of the samples. 4) A considerable percentage of both the 90-percent and average monitor sampling distributions tend to be normally distributed. 5) The distribution of 5-min occupancy samples taken consecutively on a LM channel for a period of a day (about 144 samples) usually is not similar to any common statistical distributions for low-occupancy channels. The. distribution of these samples on high-occupancy channels tends toward a normal distribution. 6) The distribution of 90-percent-occupancy samples taken by the monitor is usually asymmetric with respect to the 90-percent-occupancy value determined by use of all of the 5-min samples for a day.
Read full abstract